Friction controls earthquake physics. No direct access to the âearthquake engineâ. Earthquake simulation in the lab: friction machines. Examples of experiments ...
Earthquake fault dynamics: Insights from laboratory experiments INGV
S. Nielsen
G. Di Toro, A. Niemeijer, E. Spagnuolo, M. Violay, S. Smith, N. De Paola, A. Bistacchi, G. Pennacchioni, F. Di Felice, G. Romeo, R. Han, T. Hirose, K. Mizoguchi, M. Cocco, T. Shimamoto.
This research is funded by the European Research Council
Acknowledgements S. Nielsen (1) G. Di Toro (1, 2) A. Niemeijer (1, 4), E. Spagnuolo (1), M. Violay (1), S. Smith (1), N. De Paola (5), A. Bistacchi (6), G. Pennacchioni (2), F. Di Felice (1), G. Romeo (1), R. Han (7), T. Hirose (7), K. Mizoguchi (7), M. Cocco (1), T. Shimamoto (3), ...
(1)
(2)
(3)
(5)
INGV Geoscienze, Uni. Padova
(6)
(7)
(4)
This research is funded by the European Research Council
Earthquakes are caused by fracture and frictional slip on faults. Friction is a key parameter in understanding the physics seismic source. Let's investigate friction under Earth crust conditions.
Fault surface in dolomite, Southern Alps, Italy
Arrest or propagation of dynamic rupture
Friction law Energy balance: dissipation, heat, seismic radiation
Rupture velocity Slip velocity, rise time
Friction and energy peak
recovery
decay steady-state
L'AQUILA Courtesy of: Stefano Pucci, INGV
TOHOKU Courtesy of: Singapore Earthquake Observatory and USGS
20 km
No direct access to the “earthquake engine”...
Let's try to re-create it in the lab?
Synopsis Friction controls earthquake physics No direct access to the “earthquake engine” Earthquake simulation in the lab: friction machines Examples of experiments and typical results Lubrication processes and rock types Extrapolating the results: faults are weak during earthquakes Conclusions and future research
Experimental machines LOW STRAIN (old-style shear or rotary) HVRF (High-velocity, Rotary Shear machines) SHIVA (Slow to HIgh Velocity Apparatus) Experimental machines are NOT designed to predict earthquakes!! They help our understanding of earthquake physics and thus improve modeling and risk assessment.
µ = τ / σn Shear stress, GPa
e y “B
s ’ e rle
” w la
1.0
0.5
Low friction during EQs? 0.5
1
1.5
2.0
Normal stress, GPa [Byerlee, PAGEOPH,1978]
Rate and State
µ A
B Dc
V=0.4 mm/s V= 4 mm/s
µ friction coeff. A & B constants Dc critical slip 10-6-10-4 m θ (s) state variable V slip rate - s slip
[from Marone, 1998]
friction drop is small velocity dependance is small stress and velocity cover only part of seismic faulting conditions
The principle of ROTARY machines
Rotating shaft
HOLDERS
(torque act.)
PAIR OF CYLINDRICAL SAMPLES
Axial shaft (normal load act.)
Experimental conditions “Traditional” machines SHIVA still out of reach...
Experimental conditions and earthquake conditions
Shallow “Typical” earthquakes crustal earthquakes
Deep earthquakes
Machine “1”, Kyoto Univ., Japan, ca. 1990 (T. Shimamoto)
NIED 2007 Designed by Mizoguchi
σ n < 20 MPa v = 0.1 µm/s - 10 m/s d = infinite
SHIVA, Italy, Sept 2009 (designed by an Italian Team at INGV) σ n < 50 MPa v = 1 µm/s - 9 m/s d = infinite Seismic accelerations
1m
SHIVA, Italy, Sept 2011 INGV
1
3
2a
2b
SERVO-CONTROL 200kW drive
Video examples of the experimental dynamics
Dry, slicatic rocks --> melting
Natural faults Lab
MYLONITE (Maine)
SILTSTONE
METAGABBRO (Premosello-Italian Alps)
What do we measure?
FRICTIONAL resistance of the sample while it is sliding SHORTENING of the sample under the effect of frictional wear GAS EMISSIONS during the sliding TEMPERATURES in or around the sample close to the slip surface SLIDING VELOCITY is retrieved from rotative motion and sample radius
Example:
Synopsis Friction controls earthquake physics No direct access to the “earthquake engine” Earthquake simulation in the lab: friction machines Examples of experiments and typical results Lubrication processes and rock types Extrapolating the results: faults are weak during earthquakes Conclusions and future research
High velocity friction; lubrication
Large slip & slip-rate Intermediate normal stress Considerable WEAKENING
Lubrication effect Byerlee friction H-V friction
Nielsen et al., 2008
After Di Toro et al., 2005
The key is high work rate generating heat density
Heat density
Temperature increase
Specific heat of water, evaporation, fluid pressurization
Latent heat of chemical processes and phase transitions (decarbonation, gelification, dehydration, serpentinization, poorly known tribolchemical processes...)
Dynamic fault slip, heat and friction:
Large, fast slip means concentrated heat Heat triggers a variety of weakening mechanisms Under favourable conditions melt is produced
In its most straightforward manifestation, elevated heat density induces temperature rise and eventually yields to melting... But not always! ( maybe even seldom )
Observed processes in HV rock friction: Silica gel lubrication (quartz rocks) Thermal decomposition and nanopowders (limestones) Thermal decomposition and pressurization (dolostones) Clay-gouge weakening, dehydration and press. (clay-gouges) Flash heating (small slip amounts) Melt lubrication (all silicate built rocks)
A very inclomplete classification of rock types and frictional processes Cohesive Silicatic
Melting Silica gel lubrication Flash heating Dehydration
Carbonatic
Decarbonation Plastic yielding Dehydration Flash heating Nanopowder lub.
Non-cohesive & clays
Flash heating Decarbonation Nanopowder lub.
For most process and rock types lubrication is observed at high velocity
Not much weakening..
Friction Di Toro et al., in preparation
Seismic slip rates
Significant weakening around 0.1 m/s Different rock types (carbonates, silicates, ...)
Friction Di Toro et al., in preparation
Seismic slip rates
Work rate or power density
Friction Di Toro et al., in preparation
Cohesive Rocks: desaggregation in the results for various compositions suggest different thermal activation mechanisms.
I AT ON RB CA DE
GE LI F ICA TIO N
MELTING ON
with gouge - Brantut et al., 2008 Fault zone rich in kaolinite dehydratation at ~1m/s, ~1 MPa
Clay-clast aggregates (FE-SEM image)
(e.g., Bouteraud et al., 2008)
MELTING: glass and survivor clasts (SEM image)
pyroxene with embayment
vesicles
glass feldspar clast
20 µm
Synopsis Friction controls earthquake physics No direct access to the “earthquake engine” Earthquake simulation in the lab: friction machines Examples of experiments and typical results Lubrication processes and rock types Extrapolating the results: faults are weak during earthquakes Conclusions and future research
Extrapolation How to upscale lab to EQ conditions? Larger σn,V, accelerations Larger size than sample Complex slip time-history Complex geometry Different petrology/chemestry Different fabric
Extrapolation
I SEISMICITY, statistics of a fault population, correlations, criticality....: the process from a global point of view – YESTERDAYS TOPIC
III II LAB: small portion of a fault - scale 2 - 5 cm
SISMO: “global” rheology of the whole fault
Conclusions No direct access to the “earthquake engine” Earthquake simulation in the lab: friction machines High POWER (hich slip vel. and normal load combined) yields lubrication Various lubrication processes depending on rock types Extrapolating the results: power laws and geometry problems We barely start to measure and understand...
